Fig. 3: The association between plasma VA levels and MCI.

a Forest plot for relation between plasma VA level and the risk of MCI. b Dose-response relation between plasma VA level and the risk of MCI. c Forest plot for relation between plasma VA:(TC + TG) level and the risk of MCI. d Dose-response relation between plasma VA:(TC + TG) level and the risk of MCI. Logistic regression analysis was used to analyze the association between plasma VA or VA: (TC + TG) level and risk of MCI. Logistic regression was adjusted by age, gender, BMI, education, physical activity, living alone, reading habits, using TV and computer, smoking, drinking alcohol, usage of dietary supplements, AD family history, hyperlipidemia (not adjusted for the plasma VA/(TC + TG) level), stroke, chronic kidney disease, and T2DM in model 1. Model 2 was further adjusted by LBS and HBS in model 1. Model 3 was further adjusted by DQD in model 1. Restricted cubic spline analysis was used to analyze the dose-response relationships with four knots at the 20th, 40th, 60th, and 80th percentiles of plasma VA or VA/(TC + TG) levels, adjusted by covariates in model 2. In Fig. 4a, c, the red, purple, and green dots were represented of model 1, model 2, and model 3, respectively. In Fig. 4b, d, the four vertical lines indicated the 20th, 40th, 60th, and 80th percentiles of plasma VA or VA/(TC + TG) levels, respectively.